Why even a moderate structural realism largely eludes to be concerned with statistical thermodynamics